Information identified as archived is provided for reference, research or recordkeeping purposes. It is not subject to the Government of Canada Web Standards and has not been altered or updated since it was archived. Please "contact us" to request a format other than those available.
Concepts, methods and data quality
Data sources and analytical techniques
Most of the data in this analysis are from the 2004 Canadian Community Health Survey (CCHS): Nutrition, which was designed to collect information about the food and nutrient intakes of Canadians /concepts/hs-es/index-eng.htm). The CCHS excludes members of the regular Canadian Forces and people living in the territories, on Indian reserves, in institutions, in some remote regions, and all residents (military and civilian) of Canadian Forces bases. Detailed descriptions of the CCHS design, sample and interview procedures are available in a published report (Béland 2002).
A total of 35,107 people completed an initial 24-hour dietary recall. A subsample of 10,786 completed a second recall three to ten days later. A five-step method was used to maximize recollection of food consumed the previous day:
The response rate for the first interview was 76.5%. The response rate for the second interview was 72.8%. Composition of the food in terms of macro- and micronutrients came from the Canadian Nutrient file 2001b Supplement of Health Canada.
A total of 112 cases with invalid intake and 20 cases with null intake were excluded from this analysis. Pregnant women (175), women who were breastfeeding (91), and 4-year-old children who were being breastfed (3) were also excluded.
Data collected on the first interview day were used to estimate, by selected characteristics, average energy intake (calories) and average percentages of energy from fat, protein and carbohydrates. To determine the calories derived from each of these three macronutrients, amounts in grams were multiplied by 9, 4 and 4, respectively. Averages were defined as the average of the ratios for each individual. Energy intake includes calories from alcoholic beverages (7 calories per gram), but the percentage of energy coming from alcohol is not presented separately.
Data from both interview days were used to estimate usual intake from the macronutrients using the Software for Intake Distribution Estimation (SIDE) program (Novenario 1996; Dodd 1996) (see Appendix B One-day versus usual intake).
The foods (basic food items, recipes, or ingredients) were categorized into four groups as defined in Canada’s Food Guide to Healthy Eating (Health Canada 1997) —vegetables and fruit, milk products, meat and alternatives, and grain products—and an “other foods” category. There was no double-counting; for example, if a recipe was coded as “other foods,” the recipe, not the ingredients, was used, and vice versa. As with the macronutrients, descriptive statistics were used to estimate daily calories from each food group and the number of servings consumed per day. The distribution of usual servings from each food group was estimated with the SIDE program.
Quantities expressed in grams were transformed into servings for vegetables and fruit, milk products, and grain products, using the Canadian Nutrient File (Health Canada 2005) provided by Health Canada. Quantities for the meat and alternatives group were expressed in terms of cooked meat, with one serving containing 50 to 100 grams of meat. Servings defined without a range (peanut butter, for example) were multiplied by a factor equal to 50 grams of cooked meat.
The percentage of energy derived from a particular grouping of foods was defined as total calories from that grouping in a population, divided by total calories consumed by that population. The same methodology was used to calculate the percentage of fat coming from particular groupings of foods.
In Table 4, the foods accounting for the most calories from “other foods” were derived using food item and recipe categories. Categories are specific to a food item or a recipe. Some categories are similar for food items and recipes. Therefore, salad dressings and fruit drinks include elements assigned as a food item or as a recipe.
In Table 6, which shows the foods accounting for the most fat consumed in a day, basic food items and recipes were considered. The categories “sweet baked goods,” “milk and milk-based beverages,” “chicken dishes” and “egg dishes” are from food and recipe categories. However, “salads” include dressing only if it is part of the recipe, not if it is reported separately. “Pasta dishes” do not include pasta reported separately, and “cheese dishes” do not include cheese reported separately.
In Tables 4 and 6, the percentage of energy or fat is defined as total calories or total fat from a category, divided by total calories or total fat for all categories. The result is a picture of the population, not of an individual.
The percentage of the population who had a specific meal (breakfast, lunch, dinner) or ate between meals (snacks) was defined as the number of people who did so the first day of the interview divided by the total population reporting on the first day. This percentage is a snapshot of a given day. Therefore, it does not show the frequency with which individuals have a particular meal or consume snacks over time. The percentage of calories from a meal was also defined as the number of calories that the overall population consumed from that meal, divided by the total number of calories the population consumed in a day.
The same methodology was used to determine locations where food was prepared (home, fast-food, other). Again, these figures represent a given day, not the behaviour of any particular individual.
It is possible that the results for adults based on household income would differ if age were taken into account, because household income varies considerably by age. However, even when the data are age-adjusted, the relationships between nutritional intake and household income persist.
The bootstrap method, which takes account of the complex survey design,(Rao, Wu, Yue 1992; Rust, Rao 1996; Yeo, Mantel, Liu 1999) was used to estimate standard errors, coefficients of variation and confidence intervals. The significance level was set at p < 0.05.
Published results from the 1970-1972 Nutrition Canada Survey were used to compare the energy intake of Canadians three decades ago with results for 2004. The response rate for the 1970-1972 survey, which collected data for 10,994 respondents aged 5 or older, was 47%.
One-day versus usual intake
Two food consumption concepts must be distinguished: one-day intake and usual intake. One-day intake is total nutrient intake over a specific 24-hour period. These data are collected during an interview in which respondents are asked to describe everything they ate from midnight to midnight the previous day. Usual intake is an overview of food typically consumed in a day.
Usual intake cannot be directly estimated based on one-day intake. Usual intake varies from one individual to another. One-day intake, too, varies from one individual to another, but for a given individual, it also varies from day to day. Therefore, to estimate usual intake, it is necessary to separate the variation of an individual’s intake and that between individuals. To do this, Software for Intake Distribution Estimation (SIDE), developed by Iowa State University (Novenario 1996), was used. With a series of mathematical transformations, the software is able to estimate each component of the variance and to estimate the distribution of usual intake of a nutrient (Dodd 1996; Nusser, Carriquiry, Dodd et al. 1996). For these calculations, the day of the week was used to partially account for the effect of classification. Since the average from the first interview day is used as a benchmark, averages using one-day intake and usual intake are the same. However, estimates of the proportion of the population below or above a given threshold require a usual intake distribution.
The two curves in Chart 17 show the distributions of the percentage of calories from fat for one-day intake and usual intake for the population aged 19 or older. Average intake is essentially the same in both cases, but the distributions differ radically. Usual intake varies much less than one-day intake because, to estimate usual intake, day-to-day variation for individuals was removed.
Respondents to the 2004 Canadian Community Health Survey (CCHS): Nutrition were asked where the food they ate had been prepared: home, which includes someone else’s home; fast food, which includes take-outs and pizzerias; and other locations. Other locations cover: restaurants with waiter/waitress; other restaurants; bars, taverns, lounges; school and non-school cafeterias; work; child care centres; family/adult care centres; vending machines; grocery stores; corner stores; other stores; and other locations. The categories used in this analysis are: home only, at least some fast food (fast food only; fast food and home; fast food and other; fast food, home and other); and other combinations. When responding to the question, some respondents may have provided information about the location where they consumed the food rather than the place where it had been prepared. If a respondent reported having eaten in a fast-food establishment, he or she was considered to have eaten food prepared in a fast-food restaurant on the interview day.
For each food that they had eaten, respondents specified occasion: breakfast, lunch, dinner and between-meal consumption. Breakfast includes brunch. Between-meal consumption covers anything that was not reported as breakfast, lunch or dinner. It includes snacks, drinks consumed outside of a meal, extended consumption (eating or drinking something throughout the day), and other unspecified occasions.
For ease of reference, the term “calorie” is used in the text, although the scientifically accurate term is “kilocalorie.
Age groups were defined according to the dietary reference intake groups used by the Institute of Medicine (IOM): 4 to 8, 9 to 13, 14 to 18, 19 to 30, 31 to 50, 51 to 70, and 71 or older. Except in Table 2, data on milk products were presented for the 4 to 9, 10 to 16, and 17 or older age groups, which are used by the Canadian Food Guide "to Healthy Eating for People Four Years and Over" to establish guidelines for daily servings. To be comparable with the other food groups, the IOM age breakdowns are used for milk products in Table 2 .
Household income was calculated based on the number of people in the household and total income from all sources in the 12 months before the Canadian Community Health Survey (CCHS) interview:
In the charts, the two lowest income groups were combined.
Since the data are self-reported, respondents may not recall exactly what they ate or how much. To minimize recall errors, the 2004 Canadian Community Health Survey (CCHS): Nutrition used the five-step multiple-pass method (Moshfegh, Borrud, Perloff 1999; Moshfegh, Raper, Ingwersen et al. 2001) which was developed in the United States. Under controlled conditions, this method has effectively assessed average energy intake (Conway, Ingwersen, Vinyard 2003; Conway, Ingwersen, Moshfegh 2004), but in different settings, some studies show underreporting (Johnson, Soultanakis, Matthews 1998; Jonnalagadda, Wikman, Ahren et al. 2001; Johanssen, Wikman, Arhen et al. 2001) and others, overreporting (Gersovitz, Madden, Smicklas-Wright 1978; Myers, Klesges, Eck et al. 1988; Kahn, Appel et al. 1995).
The fact that eating occasions were self-defined may affect the results. For instance, respondents’ definitions of breakfast may range from as little as a cup of coffee to a full meal, and a snack could be a 400-calorie muffin or a cup of tea without milk or sugar. Such variations influence the percentage of calories consumed at different occasions.
Analyses based on calories (for example, percentage of calories) may not allow the role of vegetables and fruit or of “other foods” to be assessed appropriately. Vegetables and fruit tend to be low in calories, so while they contribute relatively few calories, they may be the source of a large number of servings. On the other hand, because “other foods” are often high in calories, the high percentage of calories coming from this category might reflect relatively few servings.
Despite efforts to ensure an equitable representation of days of the week during data collection, some days could be under-represented, and a bias is always possible.
The data on occasion (breakfast, lunch, dinner or snack) and location (where food was prepared) present a snapshot of a given day. These data should not be interpreted as behaviour of specific individuals.
Children younger than 6 did not reply directly to the 2004 CCHS; a parent responded on their behalf. However, a parent may not know exactly what a child ate when they were not together (at a daycare, for instance).
No statistical comparisons were made between the 2004 CCHS and the 1970-1972 Nutrition Canada Survey; the estimates for 1970-1972 in this article are based on a published report. Moreover, some concepts and collection methods differ between the two surveys. In 1970-1972, collection was done manually by dieticians/nutritionists, whereas in 2004, trained interviewers used an automated system. As well, the 1970-1972 response rate (47%) was much lower than that obtained in 2004 (77%).
For more details on the limitations of the survey, see The Canadian Community Health Survey 2.2, Nutrition Focus: A Guide to Accessing and Interpreting the Data, published by Health Canada and available on its Web site ( http://www.hc-sc.gc.ca/fn-an/surveill/nutrition/commun/index_e.html).